Association between diabetes mellitus and pancreatic cancer: A propensity score-matched cohort study using the TriNetX network

A BSTRACT Objectives: Diabetes mellitus (DM) is common and has been suspected to influence cancer risk. This study aimed to investigate the association between DM and pancreatic cancer among patients aged 20–84 years. Materials and Methods: We conducted a retrospective cohort study using the TriNetX Asia-Pacific Collaborative Network (2007–2025). Patients aged 20–84 years were divided into two groups: Those with a diagnosis of DM (DM group) and those without (control group). Propensity score matching was applied to balance baseline characteristics. The primary outcome was the incidence of pancreatic cancer. Cumulative incidence, hazard ratios (HRs), and 95% confidence intervals (CIs) were calculated using the TriNetX “Compare Outcomes” module. To mitigate potential reverse causality, we applied prespecified lag periods (primary analysis: 1-year lag; secondary analyses: 1-month, 3-month, and 6-month lags). Results: After matching, each group included 511,344 patients. Over the observation period (maximum up to 20 years), the cumulative incidence of pancreatic cancer was 0.045% in the DM group and 0.006% in the control group in the 1-year lag primary analysis. The Cox model showed that DM was associated with a higher hazard of pancreatic cancer (HR 4.677; 95% CI: 3.214–6.807). Conclusion: DM was associated with a higher observed hazard of pancreatic cancer in this large real-world cohort. However, the absolute risk was low, and findings should be interpreted considering potential detection bias and under-ascertainment of confounders in federated electronic health record data.